Michael Fischbach
Liu (Liao) Family Professor of Bioengineering, ChEM-H @Stanford.
- Today we report that an engineered skin bacterium, swabbed gently on the head of a mouse, can unleash a potent antibody response against a pathogen. Could lead to topical vaccines that are applied in a cream. @djenetbousbaine.bsky.social led the charge... @natureportfolio.bsky.social 1/55
-
View full threadWe couldn't have done this without generous financial support from Open Philanthropy, @gatesfoundation.bsky.social, @helmsleytrust.bsky.social, @su2c.bsky.social, @czbiohub.bsky.social, and NIH, and incredible institutional support from @stanford-chemh.bsky.social, IMA, and Stanford BioE. 54/55
- Finally, we are deeply grateful to Mark Leslie, @dpessential.bsky.social, Solina Chau, and Marc Benioff for supporting this work via the Microbiome Therapies Initiative. 55/55
- A few notes of gratitude. First, many thanks to @brucegoldman for explaining this so clearly... 52/55 med.stanford.edu/news/all-new...
- Christina Tobin Kåhrström and our referees were kind and constructive. Our paper was greatly improved by their suggestions. 🙏 53/55
- Fourth, is there a way to access this topical route of entry without the bacterial cell? If we understand the mechanism of uptake, can we hitch a ride using purified components applied topically? 50/55
- Finally, the level of antibody against S. epi in humans is striking. Are the human T and B cell repertoires full of clones specific for microbial colonists? 51/55
- Second, how does skin colonization induce IgA in the pulmonary and nasal mucosa? Is it due to incidental nasal colonization or do IgA-producing B cells travel from the skin to mucosal surfaces? Does the response spread to the vaginal, gut & rectal mucosa? 48/55
- Third, can this approach be optimized and extended to other pathogens? More on this soon. 49/55
- 4) Most vaccines are administered by intramuscular injection. S. epi is topical; it could be formulated in a cream and administered without a needle or a healthcare worker, which could facilitate low-cost distribution. 46/55
- There are several open questions. First, what is the mechanism of sampling? What exactly is captured (whole cell or a fragment), by which APC(s), and what is the relay that transmits antigen to GCs and tertiary lymphoid structures? 47/55
- 2) Conventional = one-time bolus w/ subsequent booster doses. Colonization is fundamentally different; antigen is encountered in a slow/steady fashion over weeks. Conventional vaccines delivered slow/steady yield better and broader responses. 44/55 www.cell.com/cell/pdf/S00...
- 3) Most traditional vaccines require a barrier breach and result in an inflammatory response (i.e., fever and swelling). Vaccination with S. epidermidis is topical and does not induce inflammation. 45/55
- Second, recall that we observed a very high level of antibody against S. epi in human serum. This makes us think that engineered S. epi could be therapeutically relevant - a new kind of vaccine that is different in four ways: 42/55
- 1) Protein-based subunit vaccines are more immunogenic when multivalent (e.g., nanoparticle-based), facilitating a more robust B cell response. S. epi displaying an Aap-scaffolded epitope is, in essence, a particulate immunogen with much higher valency. 43/55
- OK, so what did we learn? Let's start w/ the initial question: What does the immune system intend to do when it sees a colonist? Don't want to over-generalize, but at least for S. epi, the immune response is much more pathogen-like that we had realized. 40/55
- Major distinguishing feature: it is *pre-emptive*. The host reaches outside, grabs bacteria (or bacterial fragments), brings them inside, & develops an immune response against them. In a very real sense, the host is *vaccinating itself* against the colonist. 41/55
- Conventional vaccines (administered i.m.) only elicit antibodies in circulation. There is a great need for vaccines that also elicit antibodies in the lungs and nostrils - these would be particularly helpful to block transmission of respiratory viruses. 38/55 www.nature.com/articles/s41...
- Aap-sc-TTFC & the conventional vaccine both elicit a potent response in the bloodstream. However, when we looked in the pulmonary & nasal mucosa, we found that Aap-sc-TTFC had elicited a strong IgA response; the conventional nanoparticle vaccine had not. 39/55
- To our surprise, Aap-sc-TTFC elicited a massive response - even stronger than the original genetically encoded (Aap-TTFC) strain, and almost as potent as the hottest conventional vaccine we could find. Titers from engineered S. epi are stronger than a typical mRNA vaccine. 36/55
- But there is more. Antibodies exist mainly in two places: in circulation, where they protect against pathogens in the bloodstream; and in the mucosal surfaces of the lung and nostrils, where they protect against pathogen entry. 37/55
- Djenet conjugated recombinant TTFC-SpyTag to the Aap-SpyCatcher strain and colonized mice (we refer to this strain as Aap-sc-TTFC). She compared this to two other test articles: 34/55
- 1) The original (genetically encoded) strain, Aap-TTFC, applied topically; and 2) an extremely potent nanoparticle vaccine w/ the same immunogen (Mi3-TTFC), administered intramuscularly w/ adjuvant. 35/55
- Djenet engineered S. epi to express a variant of Aap that displays SpyCatcher. When she incubated these cells with recombinant GFP-SpyTag, not only were the cells labeled but the copy number was very high (we estimate ~40,000-50,000/cell). 32/55
- At this point I got on board. Transient contact with antigen can be enough to yield a long-lasting B cell response. Maybe this impermanent labeling approach might work after all, if only weakly? 33/55
- Djenet (wisely) ignored my advice and decided to do it anyway. Having received no help from me, she did what everyone in my lab does when they run into trouble: walked downstairs to consult @cobarnes27.bsky.social. 30/55
- Christopher had a clever idea. He had been making nanoparticle vaccines in which the particle bears a SpyCatcher domain, the immunogen is SpyTagged, and the two 'click' together spontaneously. Why not express SpyCatcher on the surface of S. epi? 31/55
- This was exciting, but Djenet pointed out that our approach wouldn't work with immunogens that aren't folded or modified properly when expressed in a bacterial cell - most notably, large viral proteins that are heavily glycosylated. 28/55
- She proposed an alternative: make the immunogen in mammalian cells and then attach it (after the fact) to S. epi. I dismissed the idea, arguing that the bacterial cells would divide, the immunogen would dilute, and the antibody response would be weak. Seemed inelegant. 29/55
- ...which were (more than) sufficient to protect against a lethal challenge by tetanus toxin. 26/55
- This experiment was the moment we realized how powerful this response is. Remember: all that happened to the mice was a gentle swabbing on the head. Quite a potent physiological response from such a simple intervention. 27/55
- She tried a few different strategies. The one that worked best was to replace the parallel beta-helix domain with our model antigen, a C-terminal fragment from tetanus toxin (TTFC) that is non-toxic but can elicit immunity against tetanus. 24/55
- When she colonized mice with an engineered strain of S. epi displaying TTFC within Aap, the mice developed very high titers of antibody against TTFC... 25/55
- There is much to say about (& do with) these giant repeat proteins. For now, I will focus on the perplexing fact that repeat sequences drive B cells crazy. It would appear that S. epi is goading the immune system into developing a strong antibody response against it. 22/55
- Having seen that the antibody response targets a giant cell surface protein, Djenet wondered: what if we express a non-native immunogen on the cell surface? Will the mice develop immunity against the new antigen? 23/55
- Aap is a giant (140 kDa) tree-like protein whose rigid 'trunk' (the B domain) sticks straight up through the peptidoglycan meshwork, displaying a parallel beta-helix domain that extends above the arbor on the bacterial surface. 20/55
- It consists entirely of repeat sequences. The B domain (& blue N-terminal domain) are direct amino acid repeats, while the beta-solenoid is a structural repeat. All of these presumably fold cooperatively after snaking through the secretion pore. 21/55
- Indeed there was: a mutant missing the gene that encodes sortase, the enzyme responsible for attaching proteins to the cell wall of Gram-positive bacteria. This made things easy - sortase substrates can be ID'd computationally by their C-terminal LPXTG tag. 18/55
- There are only 9 predicted sortase substrates in the two S. epi isolates we used. Djenet narrowed it down to one that is the predominant (though not exclusive) target of the antibody response: a truly bizarre protein named Aap. 19/55
- Next, Djenet wanted to figure out: what molecule on the bacterial cell surface is the target of the antibody response? 16/55
- She took a series of S. epi mutants, each one missing a conserved structure on the cell surface. By incubating them with serum from a mouse colonized with the wild-type strain, she asked: is there a mutant to which the antibodies no longer bind? 17/55
- ...presumably because they sample the hair follicle and transport bacterial antigens to the inside. Suggests that the ability of S. epi to elicit a potent immune response is not magical or inexplicable - it is a highly orchestrated physiologic process. 14/55 www.nature.com/articles/s41...
- Our paper went in a different direction. First, Djenet asked whether the antibody response to S. epi is conserved in humans. She found that serum samples from healthy adults harbor very high levels of IgG against common strains of S. epi. More on this below. 15/55
- Shortly thereafter, we learned that Inta Gribonika in the Belkaid lab had made a similar observation (neither of us realized we were both working on it!). Yasmine is one of my dearest collaborators; we coordinated our efforts from that point forward. 12/55
- They report that S. epi induces a tertiary lymphoid structure near the site of colonization; antibodies are made in this structure and in conventional germinal centers in the draining lymph node. Langerhans cells are required for all of this to happen... 13/55
- The experiment was very simple. We didn't prep the mice in any way; Djenet dipped a Q-tip in a bacterial culture and gently swabbed the head of the mouse. The result was striking: not only is there a systemic antibody response, but it is extremely potent. 10/55
- Unlike a normal antibody response following vaccination - which rises after the prime, falls back down, and rises again after the boost - the antibody response to S. epi keeps rising and rising until ~6 weeks... and then stays there indefinitely. 11/55
- Djenet grew suspicious that the immune response against this (generally harmless) commensal might be more aggressive than we had realized. When the adaptive immune system means business, both arms are typically engaged. 8/55
- So she decided to test whether, in addition to T cells, there was a B cell response as well. 9/55
- The central actor here is the ubiquitous skin colonist Staphylococcus epidermidis (S. epi). >10 years ago, @BelkaidLab discovered that the immune system mounts a peculiar T cell response against S. epi that was thought to maintain homeostasis. 6/55 www.science.org/doi/10.1126/...
- Last year, we reported (with @yerinchen.bsky.social in the lead) that the T cells elicited by S. epi are surprisingly aggressive - capable of eliminating a difficult tumor far from the site of colonization. 7/55 www.science.org/doi/10.1126/...
